Small Unmanned Helicopter Vision-guided Landing Technology Research

With the development of science and technology, the vision measurement technology of small unmanned aerial vehicle(UAV) has been widely used in many fields.As the autonomous landing of small Unmanned Aerial Vehicle is an important part during the course of the flight, landing precisely and smoothly can ensure the sections of UAV in good condition and prolong the useful life of UAV. In the past, primarily navigation system and positioning system are applied in the UAV attitude measurement and position and attitude solution is the method of the measurement of UAV. But this type of attitude measuring method has a poor anti-interference ability and it could not achieve the position and attitude calculation and the autonomous landing very well.Aiming at the shortcoming which may exist in the getting information of positioning system, this paper made a further study of algorithm of the small unmanned helicopter based on monocular visual pose measurement to get attitude information of UAV and make preparations for application on visual navigation technology of UAV guidance landing. Firstly, this paper set up a hardware system suitable for small unmanned helicopter visual measurement, which consist of a CCD camera image-collecting module, PC104 of computer module and power supply modules, and then during the flight of UAV the cooperative target information on the ground was captured by the CCD camera. Secondly, the computer installed on the UAV in order to gather the cooperative target information image in the preprocessing step and used the Harris corner detection algorithm to extract information of the cooperative target. Using the similar iterative algorithm and the two-stage iterative algorithm to solve the target attitude of the UAV and we choose the better one to solve the UAV relative to the cooperative target attitude by comparing the both algorithms. Finally a vision-based tracing scheme for landing of small Unmanned Aerial Vehicle was designed. This studyfocuses on the vision-based two-stage iterative algorithm by solving the absolute position value and the depth of field the both iteration steps. Compared to the traditional Pn P algorithm, it makes it a better noise robustness and has the advantage of rapid calculation speed and less computation, high location accuracy, high speed of convergence and small convergence error, etc. It is further validated that the two-stage iterative algorithm with high robustness is correct and effective through the experimental simulation.